Hi guys,

We have an existing Exchange deployment running 2013 on physical tin. There is a single DAG, running on 4 servers. We have 2 servers in each geographical site, with a fast WAN connection between the two sites. I'm using a shared namespace, and I'm using a clustered ARR at each site for my reverse proxy. All client traffic traverses the reverse proxy, so if you access outlook.domain.com you hit the ARR VIP, and that then makes a request to the most appropriate Exchange server.

For the Lync deployment, I have two Enterprise Pools. Again, one pool at each site. Each pool will have active users, split about 50/50, and the pools are paired. So in the event of a failover event all users will be moved from Pool A (in Site A) over to Pool B (that's in Site B).

I'm at the point where I'm integrating Lync IM and presence into OWA. Most of the config is done, I just need to edit the web.config files on Exchange... which is where I'm a little confused! :)

TechNet and most online documentation I can find simply states that you just need to add the following to the AppSettings section in the web.config file:

<add key="IMCertificateThumbprint" value="0993022B09207E20B9416B270EAD2D651A46B447"/> <add key="IMServerName" value="frontendpool.domain.com"/>

That's fine - no problem. Done is recently for someone that has Standard Edition (e.g. a single server).
However I have two pools, frontendpool-sitea.domain.com and frontendpool-siteb.domain.com. What one shall I use? I see the possible scenarios:

• For Exchange servers in Site A, I use the front end pool in Site A. I then use the front end pool in Site B for the Exchange servers hosted in Site B.... Makes sense...!
• I can use just one FE pool name for all of the Exchange servers.... Which would mean traffic could be going over the WAN which isn't ideal from an architecture point of view.
• I could have two IMServerName sections, therefore listing both FE pool names for both sites.

I'm not even sure that I could even do the last one, I'm not sure if web.config would support having two values. And even if it did, whilst it would give the greatest resilience (I assume..?) I don't know if it would use both values equally (e.g. a basic round robin approach), or if it would simply take the first value only and just use that... If the latter then hopefully if there is a timeout or not a valid response from the first value OWA would use the second entry.
That being the case, then I have my answer - for the Exchange servers in Site A, have the FE pool in Site A as the first entry, and the FE pool in Site B as the second entry. Obviously do the opposite for Site B.

What does everyone else do in this situation? Assuming the latter option in above won't work / doesn't work as I would like it to, I'm guessing the lesser of the two remaining options would be the first choice - being to have Site A FE pool used for Exchange servers in Site A, and the opposite for Site B. This would mean that should Lync fail in Site A, but Exchange is still running you'd have a 50% chance of Lync in OWA not working (as you have a 50% chance of hitting the ARR VIP in Site A, which would try to send you to the Site A Exchange, which would then fail to load Lync IM as the FE Pool listed in the web.config on those Exchange servers is down....).

Ideas, thoughs and suggestions welcomed!

Thanks - Steve

Hi Steve

Even i think that the latter option wouldn't work and you can go only with your first option

Lets wait for some more inputs on this post